Effects of Sampling Technique on Precision Estimates for Water Quality Variables in Fish Culture Ponds

Abstract

Three sampling experiments were conducted to assess variation in water quality variables in ponds: samples from individual ponds on single dates; samples from individual ponds on different dates during a growing season; and samples from replicate ponds. Data on dissolved oxygen, temperature, Secchi disk visibility, total hardness, total particulate matter, filtrable orthophosphate, and pH exhibited wide variation and covered ranges generally expected for fish ponds. A two-stage nested (hierarchical) linear statistical model was used to estimate components of variance for the three experiments. Pond size (0.04 to 10.32 ha) and management scheme (fertilized ponds and ponds receiving daily applications of fish feed) had no influence on variation. In general, viability of water quality components increased in the order: samples within ponds, among sampling dates, and between replicate ponds. Variation was greatest for total particulate matter and filtrable orthophosphate, intermediate for dissolved oxygen, Secchi disk visibility, and total hardness, and lowest for pH and temperature. The components of variance were used to estimate expected standard errors of pond means, seasonal means, and replicate pond means for all water quality variables. Figures prepared from these estimates may be used to determine the numbers of samples, sampling dates, or replicate ponds necessary to achieve a given level of precision.